Agent-empowered Tenants

I've seen quite a few mentions in LocalGov circles around the rise in AI-assisted citizen requests, e.g. subject access requests and complaints in localgov. e.g.

I've heard anecdotally that Social Housing orgs are seeing more things like complaints that look to have had a some help from chatgpt/claude/copilot/gemini, but I've not heard about there being a massive increase in volume, yet.

This got me thinking about the following things:

1) What is next for citizens/tenants? mainstream adoption of LLM powered chat interfaces happened a while ago. Early adopters more recently have moved onto Agentic (e.g. OpenClaw launched in Nov 2025, Claude Cowork in Jan 2026). So it makes sense that for social housing the next wave of mainstream adoption will be widespread tenant use of Agentic AI.

2) What does the impending mainstream adoption of agentic AI mean for social housing organisations?

Where are we now?

Historically the use of 'agents' (in the loosest sense of the term) within social housing has been driven by the organisations themselves, prior to this current wave of AI that usually meant a really basic chatbot stuck onto the organisations website, with basic NLP, probably pulling from a limited (and possibly up to date and accurate) knowledge base, and/or with a narrow and predetermined set of use cases, and limited integration into the organisations core platforms. The main outcome from these chatbots was essentially 'call centre deflection', where a tenant may or may not get their needs met (if the need is sufficiently basic), but more often than not just served as some digital treacle for the tenant to wade through before speaking to someone who could actually help (so ultimately demand deferral vs deflection).

More recently, Housing Associations that have updated their chatbots for this modern AI era have seen some increments in value for tenants, with companies like Fuzzlab and AskPorter providing more mature, integrated solutions that enable more complex needs to be met. This is all well and good, but I don't think the paradigm of 'call centre deflection' and 'automating low value interactions so the contact centre people can deal with complex and value add' is going to be around much longer.

Skating to where the puck is going to be

When (not if) tenant use of Agentic AI becomes mainstream, the mental model through which we think about customer demand, channel shift, etc will need to change rapidly.

Imagine being a tenant and having an agent that:

Knows you and your specific context
Knows your housing association, its policies, procedures, contact channels, its organisational structure and governance, its regulatory ratings and weaknesses, local counsellors and MPs, relevant case law and recent ombudsmen findings
Knows the regulatory and legal context that your housing association works within better than 99% of the colleagues working in the organisation
Is able to tirelessly research, draft, submit, chase, escalate, negotiate, and iterate on your behalf

Now imagine being a tenant and having to raise a complaint, flag an ASB issue, Negotiate an arrears plan, or have any other interaction with your housing association. How much better equipped will you be to get the right outcome for you? how much less time and frustration will it take to get the right outcome for you?

I think ultimately, in a scenario where there is mainstream adoption of Agentic AI by tenants, this will represent a significant and fundamental shift in the tenant/housing association power dynamic, that in the long term will be a great thing for tenants (once housing associations have worked out how to respond), but in the short-medium term will generate a significant amount of work and service failure for housing associations that are not ahead of this imminent, and new type of demand.